|
1
|
|
|
from __future__ import annotations |
|
|
|
|
|
|
2
|
|
|
import abc |
|
3
|
|
|
import enum |
|
4
|
|
|
import sys |
|
|
|
|
|
|
5
|
|
|
import sre_compile |
|
6
|
|
|
import re |
|
7
|
|
|
from pathlib import Path |
|
8
|
|
|
from typing import Dict, Sequence, Type, Set, Optional, Mapping |
|
9
|
|
|
from typing import Tuple as Tup |
|
10
|
|
|
|
|
11
|
|
|
from mandos.model import MandosResources, ReflectionUtils |
|
12
|
|
|
from mandos.model.chembl_api import ChemblApi |
|
13
|
|
|
from mandos.model.chembl_support.chembl_targets import TargetType, ChemblTarget |
|
14
|
|
|
from mandos.model.chembl_support.chembl_target_graphs import ( |
|
15
|
|
|
ChemblTargetGraph, |
|
16
|
|
|
TargetNode, |
|
17
|
|
|
TargetEdgeReqs, |
|
18
|
|
|
TargetRelType, |
|
19
|
|
|
) |
|
20
|
|
|
|
|
21
|
|
|
|
|
22
|
|
|
class Acceptance(enum.Enum): |
|
|
|
|
|
|
23
|
|
|
always = enum.auto() |
|
24
|
|
|
never = enum.auto() |
|
25
|
|
|
at_start = enum.auto() |
|
26
|
|
|
at_end = enum.auto() |
|
27
|
|
|
|
|
28
|
|
|
|
|
29
|
|
|
class TargetTraversalStrategy(metaclass=abc.ABCMeta): |
|
|
|
|
|
|
30
|
|
|
"""""" |
|
31
|
|
|
|
|
32
|
|
|
@classmethod |
|
33
|
|
|
@property |
|
34
|
|
|
def api(cls) -> ChemblApi: |
|
|
|
|
|
|
35
|
|
|
raise NotImplementedError() |
|
36
|
|
|
|
|
37
|
|
|
def traverse(self, target: ChemblTargetGraph) -> Sequence[ChemblTargetGraph]: |
|
|
|
|
|
|
38
|
|
|
return self.__call__(target) |
|
39
|
|
|
|
|
40
|
|
|
def __call__(self, target: ChemblTargetGraph) -> Sequence[ChemblTargetGraph]: |
|
41
|
|
|
""" |
|
42
|
|
|
|
|
43
|
|
|
Returns: |
|
44
|
|
|
|
|
45
|
|
|
""" |
|
46
|
|
|
raise NotImplementedError() |
|
47
|
|
|
|
|
48
|
|
|
|
|
49
|
|
|
class NullTargetTraversalStrategy(TargetTraversalStrategy, metaclass=abc.ABCMeta): |
|
|
|
|
|
|
50
|
|
|
"""""" |
|
51
|
|
|
|
|
52
|
|
|
@classmethod |
|
53
|
|
|
@property |
|
54
|
|
|
def api(cls) -> ChemblApi: |
|
55
|
|
|
raise NotImplementedError() |
|
56
|
|
|
|
|
57
|
|
|
def traverse(self, target: ChemblTargetGraph) -> Sequence[ChemblTargetGraph]: |
|
58
|
|
|
return self.__call__(target) |
|
59
|
|
|
|
|
60
|
|
|
def __call__(self, target: ChemblTargetGraph) -> Sequence[ChemblTargetGraph]: |
|
61
|
|
|
return [target] |
|
62
|
|
|
|
|
63
|
|
|
|
|
64
|
|
|
class StandardTargetTraversalStrategy(TargetTraversalStrategy, metaclass=abc.ABCMeta): |
|
|
|
|
|
|
65
|
|
|
"""""" |
|
66
|
|
|
|
|
67
|
|
|
@classmethod |
|
68
|
|
|
@property |
|
69
|
|
|
def edges(cls) -> Set[TargetEdgeReqs]: |
|
|
|
|
|
|
70
|
|
|
raise NotImplementedError() |
|
71
|
|
|
|
|
72
|
|
|
@classmethod |
|
73
|
|
|
@property |
|
74
|
|
|
def acceptance(cls) -> Mapping[TargetEdgeReqs, Acceptance]: |
|
|
|
|
|
|
75
|
|
|
raise NotImplementedError() |
|
76
|
|
|
|
|
77
|
|
|
def __call__(self, target: ChemblTargetGraph) -> Sequence[ChemblTarget]: |
|
78
|
|
|
found = target.traverse(self.edges) |
|
79
|
|
|
return [f.target for f in found if self.accept(f)] |
|
80
|
|
|
|
|
81
|
|
|
def accept(self, target: TargetNode) -> bool: |
|
|
|
|
|
|
82
|
|
|
if target.link_reqs is None: |
|
83
|
|
|
# typically for root nodes -- we'll have a self-loop to check, too |
|
84
|
|
|
return False |
|
85
|
|
|
acceptance_type = self.acceptance[target.link_reqs] |
|
86
|
|
|
return ( |
|
87
|
|
|
acceptance_type is Acceptance.always |
|
88
|
|
|
or (acceptance_type is Acceptance.at_start and target.is_start) |
|
89
|
|
|
or (acceptance_type is Acceptance.at_end and target.is_end) |
|
90
|
|
|
) |
|
91
|
|
|
|
|
92
|
|
|
|
|
93
|
|
|
class StandardStrategyParser: |
|
|
|
|
|
|
94
|
|
|
@classmethod |
|
95
|
|
|
def read_lines(cls, path: Path) -> Sequence[str]: |
|
|
|
|
|
|
96
|
|
|
return [ |
|
97
|
|
|
line |
|
98
|
|
|
for line in path.read_text(encoding="utf8").splitlines() |
|
99
|
|
|
if not line.startswith("#") and len(line.strip()) > 0 |
|
100
|
|
|
] |
|
101
|
|
|
|
|
102
|
|
|
@classmethod |
|
103
|
|
|
def parse( |
|
|
|
|
|
|
104
|
|
|
cls, lines: Sequence[str] |
|
|
|
|
|
|
105
|
|
|
) -> Tup[Set[TargetEdgeReqs], Mapping[TargetEdgeReqs, Acceptance]]: |
|
106
|
|
|
pat_type = r"(@?[a-z_]+)" |
|
107
|
|
|
pat_rel = r"([<>~=.*])" |
|
108
|
|
|
pat_accept = r"(?:accept:([\-*^$]))?" |
|
109
|
|
|
pat_src_words = r"(?:src:'''(.+?)''')?" |
|
110
|
|
|
pat_dest_words = r"(?:dest:'''(.+?)''')?" |
|
111
|
|
|
comment = r"(?:#(.*))?" |
|
112
|
|
|
pat = f"^ *{pat_type} *{pat_rel} *{pat_type} *{pat_accept} *{pat_src_words} *{pat_dest_words} *{comment} *$" |
|
|
|
|
|
|
113
|
|
|
pat = re.compile(pat) |
|
114
|
|
|
to_rel = { |
|
115
|
|
|
">": TargetRelType.superset_of, |
|
116
|
|
|
"<": TargetRelType.subset_of, |
|
117
|
|
|
"~": TargetRelType.overlaps_with, |
|
118
|
|
|
"=": TargetRelType.equivalent_to, |
|
119
|
|
|
"*": TargetRelType.any_link, |
|
120
|
|
|
".": TargetRelType.self_link, |
|
121
|
|
|
} |
|
122
|
|
|
to_accept = { |
|
123
|
|
|
"*": Acceptance.always, |
|
124
|
|
|
"-": Acceptance.never, |
|
125
|
|
|
"^": Acceptance.at_start, |
|
126
|
|
|
"$": Acceptance.at_end, |
|
127
|
|
|
} |
|
128
|
|
|
edges = set() |
|
129
|
|
|
edge_to_acceptance: Dict[TargetEdgeReqs, Acceptance] = {} |
|
130
|
|
|
for line in lines: |
|
131
|
|
|
match = pat.fullmatch(line) |
|
132
|
|
|
if match is None: |
|
133
|
|
|
raise AssertionError(f"Could not parse line '{line}'") |
|
134
|
|
|
try: |
|
135
|
|
|
sources = TargetType.resolve(match.group(1)) |
|
136
|
|
|
rel = to_rel[match.group(2)] |
|
137
|
|
|
dests = TargetType.resolve(match.group(3)) |
|
138
|
|
|
accept = to_accept[match.group(4).lower()] |
|
139
|
|
|
src_pat = ( |
|
140
|
|
|
None |
|
141
|
|
|
if match.group(5) is None or match.group(5) == "" |
|
142
|
|
|
else re.compile(match.group(5)) |
|
143
|
|
|
) |
|
144
|
|
|
dest_pat = ( |
|
145
|
|
|
None |
|
146
|
|
|
if match.group(6) is None or match.group(6) == "" |
|
147
|
|
|
else re.compile(match.group(6)) |
|
148
|
|
|
) |
|
149
|
|
|
except (KeyError, TypeError, sre_compile.error): |
|
150
|
|
|
raise AssertionError(f"Could not parse line '{line}'") |
|
151
|
|
|
for source in sources: |
|
152
|
|
|
for dest in dests: |
|
153
|
|
|
edge = TargetEdgeReqs( |
|
154
|
|
|
src_type=source, |
|
155
|
|
|
src_pattern=src_pat, |
|
156
|
|
|
rel_type=rel, |
|
157
|
|
|
dest_type=dest, |
|
158
|
|
|
dest_pattern=dest_pat, |
|
159
|
|
|
) |
|
160
|
|
|
edges.add(edge) |
|
161
|
|
|
edge_to_acceptance[edge] = accept |
|
162
|
|
|
return edges, edge_to_acceptance |
|
163
|
|
|
|
|
164
|
|
|
|
|
165
|
|
|
class TargetTraversalStrategies: |
|
166
|
|
|
""" |
|
167
|
|
|
Factory. |
|
168
|
|
|
""" |
|
169
|
|
|
|
|
170
|
|
|
@classmethod |
|
171
|
|
|
def standard_strategies(cls) -> Set[str]: |
|
|
|
|
|
|
172
|
|
|
return { |
|
173
|
|
|
p.stem for p in MandosResources.path("strategies").iterdir() if p.suffix == ".strat" |
|
174
|
|
|
} |
|
175
|
|
|
|
|
176
|
|
|
@classmethod |
|
177
|
|
|
def by_name(cls, name: str, api: ChemblApi) -> TargetTraversalStrategy: |
|
|
|
|
|
|
178
|
|
|
if name == "@null": # just slightly more efficient |
|
179
|
|
|
return cls.null(api) |
|
180
|
|
|
if MandosResources.contains("strategies", name, suffix=".strat"): |
|
|
|
|
|
|
181
|
|
|
return cls.from_resource(name, api) |
|
182
|
|
|
elif name.endswith(".strat"): |
|
183
|
|
|
return cls.from_file(Path(name), api) |
|
184
|
|
|
return cls.by_classname(name, api) |
|
185
|
|
|
|
|
186
|
|
|
@classmethod |
|
187
|
|
|
def by_classname(cls, fully_qualified: str, api: ChemblApi) -> TargetTraversalStrategy: |
|
|
|
|
|
|
188
|
|
|
clazz = ReflectionUtils.injection(fully_qualified, TargetTraversalStrategy) |
|
189
|
|
|
return cls.create(clazz, api) |
|
190
|
|
|
|
|
191
|
|
|
@classmethod |
|
192
|
|
|
def from_resource(cls, name: str, api: ChemblApi) -> TargetTraversalStrategy: |
|
|
|
|
|
|
193
|
|
|
path = MandosResources.path("strategies", name, suffix=".strat") |
|
194
|
|
|
lines = StandardStrategyParser.read_lines(path) |
|
195
|
|
|
return cls._from_lines(lines, api, path.stem) |
|
196
|
|
|
|
|
197
|
|
|
@classmethod |
|
198
|
|
|
def from_file(cls, path: Path, api: ChemblApi) -> TargetTraversalStrategy: |
|
|
|
|
|
|
199
|
|
|
lines = StandardStrategyParser.read_lines(path) |
|
200
|
|
|
return cls._from_lines(lines, api, path.stem) |
|
201
|
|
|
|
|
202
|
|
|
@classmethod |
|
203
|
|
|
def from_lines( |
|
|
|
|
|
|
204
|
|
|
cls, lines: Sequence[str], api: ChemblApi, name: Optional[str] |
|
|
|
|
|
|
205
|
|
|
) -> TargetTraversalStrategy: |
|
206
|
|
|
return cls._from_lines(lines, api, "" if name is None else name) |
|
207
|
|
|
|
|
208
|
|
|
@classmethod |
|
209
|
|
|
def _from_lines( |
|
210
|
|
|
cls, lines: Sequence[str], api: ChemblApi, name: str |
|
|
|
|
|
|
211
|
|
|
) -> TargetTraversalStrategy: |
|
212
|
|
|
edges, accept = StandardStrategyParser.parse(lines) |
|
213
|
|
|
|
|
214
|
|
|
class Strategy(StandardTargetTraversalStrategy): |
|
|
|
|
|
|
215
|
|
|
@classmethod |
|
216
|
|
|
@property |
|
217
|
|
|
def edges(cls) -> Set[TargetEdgeReqs]: |
|
218
|
|
|
return edges |
|
219
|
|
|
|
|
220
|
|
|
@classmethod |
|
221
|
|
|
@property |
|
222
|
|
|
def acceptance(cls) -> Mapping[TargetEdgeReqs, Acceptance]: |
|
223
|
|
|
return accept |
|
224
|
|
|
|
|
225
|
|
|
@classmethod |
|
226
|
|
|
@property |
|
227
|
|
|
def api(cls) -> ChemblApi: |
|
228
|
|
|
return api |
|
229
|
|
|
|
|
230
|
|
|
Strategy.__name__ = StandardTargetTraversalStrategy.__class__.__name__ + "_" + name |
|
231
|
|
|
return Strategy() |
|
232
|
|
|
|
|
233
|
|
|
@classmethod |
|
234
|
|
|
def null(cls, api: ChemblApi) -> TargetTraversalStrategy: |
|
|
|
|
|
|
235
|
|
|
class NullStrategy(NullTargetTraversalStrategy): |
|
|
|
|
|
|
236
|
|
|
@classmethod |
|
237
|
|
|
@property |
|
238
|
|
|
def api(cls) -> ChemblApi: |
|
239
|
|
|
return api |
|
240
|
|
|
|
|
241
|
|
|
return NullStrategy() |
|
242
|
|
|
|
|
243
|
|
|
# noinspection PyAbstractClass |
|
244
|
|
|
@classmethod |
|
245
|
|
|
def create(cls, clz: Type[TargetTraversalStrategy], api: ChemblApi) -> TargetTraversalStrategy: |
|
246
|
|
|
""" |
|
247
|
|
|
Factory method. |
|
248
|
|
|
|
|
249
|
|
|
Args: |
|
250
|
|
|
clz: |
|
251
|
|
|
api: |
|
252
|
|
|
|
|
253
|
|
|
Returns: |
|
254
|
|
|
|
|
255
|
|
|
""" |
|
256
|
|
|
|
|
257
|
|
|
class X(clz): |
|
|
|
|
|
|
258
|
|
|
@classmethod |
|
259
|
|
|
@property |
|
260
|
|
|
def api(cls) -> ChemblApi: |
|
|
|
|
|
|
261
|
|
|
return api |
|
262
|
|
|
|
|
263
|
|
|
X.__name__ = clz.__name__ |
|
264
|
|
|
return X() |
|
265
|
|
|
|
|
266
|
|
|
|
|
267
|
|
|
__all__ = ["TargetTraversalStrategy", "TargetTraversalStrategies"] |
|
268
|
|
|
|